Scholarlink

<p>With the advent of 6G networks, the increasing demand for next-generation mobile services is bringing new challenges to network management and resource allocation. In 6G, one of the key requirements for effective service delivery is to optimize the resource allocation of Mobile Virtual Network Operators (MVNOs). In this study, we introduce an innovative decentralized convolutional autoencoder and Long Short-Term Memory (LSTM) model aimed at enhancing MVNO slicing technology in 6G networks. The proposed architecture supports the management of radio access network (RAN) slices, allowing various devices, including URLLC+, FeMBB, MEC slices, and uMTC, to send different data packets to a single base station (BS) within the uplink context. Furthermore, mobile network providers lend the physical resources of the RAN, including radio resources, to virtual mobile network operators (MVNOs), enabling them to create RAN slices tailored to specific services. Our method employs deep learning techniques to identify latent attributes from network data and accurately predict resource usage. By using a distributed learning architecture, this framework eliminates the need for centralized data storage and provides scalable and rapid training across multiple devices. Our research results indicate that, compared to traditional methods, network efficiency and the accuracy of resource allocation have significantly improved.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Distributed Convolutional Autoencoder and LSTM Model for Optimized MVNO Slicing in 6G Networks

  • 期刊论文

  • 发表日期：

    2025年9月5日

查看全文
  • Megha Jain,
  • Ravi Verma,
  • J. Amudhavel

摘要

With the advent of 6G networks, the increasing demand for next-generation mobile services is bringing new challenges to network management and resource allocation. In 6G, one of the key requirements for effective service delivery is to optimize the resource allocation of Mobile Virtual Network Operators (MVNOs). In this study, we introduce an innovative decentralized convolutional autoencoder and Long Short-Term Memory (LSTM) model aimed at enhancing MVNO slicing technology in 6G networks. The proposed architecture supports the management of radio access network (RAN) slices, allowing various devices, including URLLC+, FeMBB, MEC slices, and uMTC, to send different data packets to a single base station (BS) within the uplink context. Furthermore, mobile network providers lend the physical resources of the RAN, including radio resources, to virtual mobile network operators (MVNOs), enabling them to create RAN slices tailored to specific services. Our method employs deep learning techniques to identify latent attributes from network data and accurately predict resource usage. By using a distributed learning architecture, this framework eliminates the need for centralized data storage and provides scalable and rapid training across multiple devices. Our research results indicate that, compared to traditional methods, network efficiency and the accuracy of resource allocation have significantly improved.